QF-funded startup to ‘revolutionise’ dialectal speech recognition tech for Arabic speakers

QF-funded startup to ‘revolutionise’ dialectal speech recognition tech for Arabic speakers

An innovative ‘speech tech’ startup is aiming to lead the market in Arabic speech recognition that supports 19 dialects.

“The smaller market of Arabic speakers versus English speakers and the dialectal challenges of Arabic have limited voice technology in the region,” said Ryan Carmichael, chief executive officer of Kanari AI.

Kanari AI is a recipient of the Tech Venture Fund (TVF) from Qatar Science & Technology Park (QSTP), part of Qatar Foundation's (QF’s) Research, Development and Innovation (QF RDI). The company licensed its core technology from Qatar Computing Research Institute (QCRI), part of QF’s Hamad Bin Khalifa University (HBKU).

Carmichael said Kanari AI was launched in 2020 to provide innovative Arabic speech recognition that recognises individual dialects in audio and video files to convert them to written text.

“We created the leading dialectal Arabic speech recognition platform in the market to detect 19 Arabic dialects in one global model. This covers 95% of the market, so this really sets us apart from the competition,” he explained.

According to Carmichael, written and spoken Arabic takes many forms – from accents to linguistic inflections and regional dialects. It is assumed that there are at least 22 dialects of Arabic in circulation today, he said.

“Perhaps due to